The Sharpe Ratio does not seem to be doing its job. In the definition, which I have encluded below says that the Sharpe Ratio measures whether a trading system is like a roller coaster. It further says that its great to end up with a huge profit but its not fun if along the way you have to lose half of your money to then regain it again. In my opinion the Sharpe Ratio is not living up to its definition. All you have to do is look at the Best Systems Hot List and you will see that there are trading systems listed there with very good Sharpe Ratio’s but yet you have to lose half of your money to then have to regain it again before making a profit. I believe that this issue needs to be investigated by Matthew. Trading systems in which traders could potentially lose half their trading capital before they make a profit should not warrent such outstanding Sharpe Ratio’s.
Statistics: What is the Sharpe Ratio?
The “Sharpe Ratio” is one of the statistics Collective2 calculates for each trading system it tracks. It was developed by Nobel Laureate Bill Sharpe as a measure of the riskiness of an investment versus its profitability. The Sharpe Ratio measures whether a trading system is like a roller coaster. Sure, it’s great to end up with a huge profit, but it’s not fun if, along the way, you have to lose half of your money, then regain it, then lose it again, then regain it… The higher the Sharpe Ratio, the better.
Since a reply from MK was lacking for the past 2 days, I thought I would add my 2 cents:
At C2, the performance statistics (Sharpe Ratio etc.) are calculated from combined open and closed equity. A drawdown related statistic like Sharpe Ratio is underestimated if calculated from combined open and closed equity.
This is because, using open trade equity for drawdown calculations will overestimate the drawdown. But, in some cases, using closed trade equity for drawdown calculations will underestimate the drawdown. This is usually only a concern for long-term trend following systems (unhedged by options or futures), which may have large swings in equity during the trade. Even with these types of systems, however, only a fraction of the trades generally experience intra-trade drawdowns that completely reverse before the trade exits. If, on the other hand, a trade exits at its worst intra-trade drawdown, the closed trade equity will also accurately represent the intra-trade drawdown, a valid case for calculating the Sharpe Ratio and other relevant statistics based on the closed-equity plot.
This is a spirited debate we have had before. I’ll repeat what I said earlier. I mean this in the most constructive way: your insistence that performance statistics should not include open trades is just plain bizarre.
There is no fund manager anywhere in the entire world, no individual trader, no broker – who does not care about drawdowns for “open” trades. Performance must be marked-to-market: whether you are measuring your own performance simply for internal tracking purposes, or whether you are managing money on behalf of others. Just to point to a most obvious facet of this: brokers who hold your account will insist on your positions being marked-to-market and will restrict account buying power accordingly.
Here is a thought experiment, to show how absurd it is to claim that open positions should not be included in performance measurements. If they were not, I would simply pursue the following strategy: (1) Buy stock. (2) Hope it goes up. (3) Uh-oh. Didn’t go up? Okay, then, double position and buy more stock. (4) Hope it goes up. (5) Didn’t go up? Double position again.
According to your proposed rules, my account would not show an adverse drawdown. It would simply be a flat line until I closed the position. Under your proposed rules, I would never have any risk at all: I could simply continue to “double up” my bet until my lottery ticket finally won. Obviously this can’t be done in the real world. If it could, all sorts of “doubling-up” strategies would be tenable.
Because the government, regulators, and brokers insist on marking open positions to market, such strategies are (in the end) impossible.
Matthew
Great post Matthew !
Alan, what do you call a good sharpe ratio?
A don’t even throw an eye on systems under 1, and I would never subscribe to a system under 2.
Some systems featured as “best systems” have good returns, but low sharpe (under 1). These are junk-systems, or gambler-systems, from my point of view.
I should clarify one more thing about the Sharpe Ratio here on Collective2.
The way the Sharpe Ratio is calculated is to divide the performance record of a trading system (or portfolio, or whatever) into “periods” and then compare the excess return of each period against a benchmark (the S&P500 in this case).
All well and good. Except for one thing: You’re really supposed to calculate the track record over a period of years, and use a reasonably long periods for comparison (for example, quarters).
Collective2 takes Sharpe calculations to extremes, however. (Sort of like the “X Games” for mathematics.) It calculates Sharpe Ratios even in situations where most Finance professors would say, "Sorry, can’t do it. Not enough data."
It does this by being agnostic about what kind of periods to use in calculations. If a system has a track record of over a year or so, it uses quarterly periods. But if a system has a track record of only a few months, it might use weeks as the period for comparing returns of the system versus the S&P.
The implication here is that, while Sharpe Ratios are vaguely interesting, they are not the be-all and end-all …particularly for trading systems where there is less than a one-year track record. Take the numbers lightly.
Matthew
Here is an excerpt from adaptrade.com:
Martingale methods, decrease the amount at risk after a win and increase the amount at risk after a loss. A commonly used example is “doubling down” after a loss in gambling. Martingale methods are most often used by gamblers, who trade against the house’s advantage.
Another approach is where the number of contracts or shares increases as profits accrue and decreases as the equity drops during a drawdown. Position sizing methods that use this approach are known as antimartingale methods.
Provided you have a profitable trading method, antimartingale methods are always preferable over the long run because they’re capable of growing your trading account geometrically. However, it’s sometimes possible to lower your risk by taking advantage of patterns of wins and losses, similar to martingale methods. Two ways to do this are via dependency rules and using equity curve trading. Both of these methods can be used in addition to whichever position sizing method you choose.
While martingale methods are not recommended, these adjustments to the antimartingale methods listed above can sometimes prove beneficial. For example, if your trading system or method tends to have long winning and losing streaks, it might be beneficial to skip trading after a loss until the first skipped trade would have been a winner. Resume taking the signals until a loss is encountered. This is an example of a dependency rule for systems with positive dependency.
Alternatively, you might try trading the equity curve. For example, you could stop taking the trading signals when the moving average of the equity curve crosses below the equity curve line. Resume trading when the moving average crosses back above the equity curve.
The equity curve referred above is the closed equity curve because entries and exits are based on closed equity crossovers. That is why it would be useful to include a closed equity plot at C2 in addition to the open equity plot. This would paint a complete picture of the performance of the system. Showing only an open equity plot, paints an incomplete picture of the performance of the system.
Sorry Pal, I don’t agree with you.
Too many system developers act in an irresponsible way, here on C2.
They enter hudge trades without stop-loss, hoping that market will move in the good direction. If not, they leave their system asleep until, accidently, the losing trade turns into a win.
Indeed, this kind of system developers have strictly no skills in trading, and they are dishonest with their subscribers. They have nothing good to sell, but want to make big money, it’s a shame.
That’s why I think most systems are obviously useless on C2. The sharpe ratio is the best indicator, but only 10-20% of the systems have a decent sharpe…
There are too many gamblers on this site and Matthew must find a way to supress them, or C2 will no more be reputable.
Does C2 intend to earn money from overabundant system developers with no skills, or from happy subscribers of valuable systems? C2 have to make a choice.
C2 doesn’t need developers, it needs subscribers! But I guess there are still much more developers than subscribers. As a disappointed subscriber will never come back again, C2 shouldn’t promote low sharpe systems.
Also, I would agree with Alex Matulich who says: “Expectancy score is a better, more objective measure than the Sharpe Ratio for evaluating the relative performance of different trading strategies.” http://unicorn.us.com/trading/expectancy.html
OK, I couldn’t stop myself from responding, I apologize…
> Too many system developers act in an irresponsible way, here on C2.
That is their prerogative, I am sorry. You can not make people responsible by ordering them to do so.
>They enter huge trades without stop-loss, hoping that market will move in the good direction.
Again, it is up to them. If you don’t like the style, don’t choose the system, quite simple. Actually, you should be happy, because your system has a better chance against bad systems. 
> If not, they leave their system asleep until, accidently, the losing trade turns into a win.
Some people would call this account management. You don’t know, maybe they have a bigger timeframe, or a wider stop loss.
Again, it is up to them, you can criticize it, but the best way is to vote with your money and not to subscribe to them. It is THAT simple…
>Indeed, this kind of system developers have strictly no skills in trading, and they are dishonest with their subscribers. They have nothing good to sell, but want to make big money, it’s a shame.
You might be surprized, but not everybody here wants to get subscribers. Some of us just testing different strategies or trying out trading before using real money. You simple just don’t know the reason why a vendor started his system, unless it is stated.
>That’s why I think most systems are obviously useless on C2.
This is true, but look at it as an advantage, if you are a vendor, as you are. Would you really want to compete against 100 excellent systems with over 200% annualized gains? That would be silly.
>The sharpe ratio is the best indicator, but only 10-20% of the systems have a decent sharpe…
Just because I am in an argumentative mood this morning, my stocktrading system is up 10 %, has no downdrawn, the last 20 trades were all profitable nevertheless it has a negative Sharpe ratio, explain that to me. It did have a longer term when not traded though. The point is here that Sharpe ratio is not everything…
>There are too many gamblers on this site and Matthew must find a way to supress them, or C2 will no more be reputable.
This is the typical “I don’t take responsibility for my actions” approach. To find a good system is the subscriber’s responsibility, and not Matthew’s. If you want your hand held, get a girlfriend. Matt’s responsibility is to run the website in the possible most effective way.
Since vendors paid too for having their systems listed here and using the site, it doesn’t matter how they perform as far as the fee has been paid.
>Does C2 intend to earn money from overabundant system developers with no skills, or from happy subscribers of valuable systems?
Actually, the way how the fees are constructed right now, the answer is both. The website makes money from vendors without subscribers…
>C2 doesn’t need developers, it needs subscribers!
No. Vendors need subscribers, C2 only need paying costumers
be it vendor or subscriber…
I mean, I see your point, you want more subscribers and more traffic. I agree. But again, there are already plenty of info on the website, if a possible subscriber can not choose using that info glut, I am sorry, just throw a dart at the printed list of the C2 systems and subscribe to it.
> C2 shouldn’t promote low sharpe systems.
It doesn’t. But it lists them as it should.
This is most absurd argument I have ever seen in C2. No matter what kind of performance indicator you calculate, you must include open trades. This is just a common sense. Period.
Always nice to debate with you Peter.
I respect your opinions on that issue.
I just can’t stand seeing very poor systems on the top of the list or the so-called “best systems”. In the real life, they would have get margin calls and their accounts closed!
Of course, subscribers are supposed to be wise enough to make a better choice than this. But as C2 displays some of these systems as “top of the art”, it could be very misleading.
